Image Classification on ImageNet OOD Transfer Suite (CUB, Cars, Aircrafts, Pets) LT (test)
70.7CUB AccuracySimSiam+rwSAM
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| SimSiam+rwSAMPre-training Dataset=ImageNet-LT, Evaluation Protocol=fine-tuning, Regularization=rwSAM2021.10 | 70.7 | 88.4 | 82.6 | 84 | 81.4 | |
| SimSiam, balancedPre-training Dataset=Balanced dataset (subset of ImageNet), Evaluation Protocol=fine-tuning2021.10 | 70.5 | 87.9 | 81.8 | 82.7 | 80.7 | |
| MoCo v2+rwSAMPre-training Dataset=ImageNet-LT, Evaluation Protocol=fine-tuning, Regularization=rwSAM2021.10 | 70.3 | 88.7 | 84.9 | 81.7 | 81.4 | |
| SimSiamPre-training Dataset=ImageNet-LT, Evaluation Protocol=fine-tuning2021.10 | 70 | 87 | 81.5 | 83.8 | 80.6 | |
| MoCo v2Pre-training Dataset=ImageNet-LT, Evaluation Protocol=fine-tuning2021.10 | 69.9 | 88.4 | 82.9 | 80.1 | 80.3 | |
| MoCo v2+SAMPre-training Dataset=ImageNet-LT, Evaluation Protocol=fine-tuning, Regularization=SAM2021.10 | 69.9 | 88.8 | 83.4 | 81.5 | 80.9 | |
| MoCo v2, balancedPre-training Dataset=Balanced dataset (subset of ImageNet), Evaluation Protocol=fine-tuning2021.10 | 69.8 | 88.6 | 82.7 | 80 | 80.2 |